This proposal will examine the molecular response of cells in blood to ischemic stroke in humans. The proposal is based upon our preliminary data showing specific gene expression profiles in blood for large vessel atherosclerotic, cardioembolic and lacunar causes of stroke. These profiles were derived in part to predict the 30% of stroke patients who have unknown (cryptogenic) causes of their strokes. Using these profiles, 17% of cryptogenic stroke patients were predicted to have large vessel atherosclerosis, 41% were predicted to be cardioembolic, and 27% of the cardioembolic strokes were predicted to have paroxysmal atrial fibrillation (PAF). These findings are important since they affect what treatment is used - anticoagulants like coumadin for cardioembolic causes, vascular procedures and anti-platelet agents for large vessel atherosclerotic causes, and anti-platelet agents for lacunar stroke. Based upon these data we propose the following aims. (1) Use PCR arrays and the genes from our expression profiles for large vessel atherosclerotic, cardioembolic and lacunar stroke to predict the causes of ischemic strokes in this study using whole blood with >90% sensitivity and specificity. (2) For the cryptogenic stroke patients predicted to have cardioembolic strokes due to PAF using PCR arrays in Aim #1, demonstrate they have PAF during cardiac monitoring following their stroke. (3) Derive gene expression profiles from isolated PMNs and monocytes for large vessel, cardioembolic and lacunar causes of ischemic stroke in one half of the patients matched for age, gender, race and vascular risk factors and corrected for multiple comparisons. Show that these profiles predict the causes of ischemic stroke in the second half of the stroke patients with >90% sensitivity and specificity. The first aim will be the first to use the gene expression profiles from our previous studies of whole blood to predict the causes of stroke using PCR arrays in a separate independent cohort in this study. The second aim predicts which cryptogenic strokes are caused by PAF and will confirm these using cardiac monitoring. The studies of PMNs and monocytes will begin to assess the roles of these cells in different causes of stroke. Predicting the causes of cryptogenic strokes has the potential to soon decrease the incidence of recurrent ischemic strokes since the results can be obtained within days of the stroke using current PCR array technology. The gene profiles for cryptogenic stroke will speed delivery of the most appropriate treatments and make further evaluations of the causes of the cryptogenic strokes quicker, less expensive and more successful.